PolicyResearchW

WORKING PAPERS

TradePolicy

Public Disclosure Authorized CountryEconomics Department TheWorld Bank February1993 WPS1094

The New Regionalism

Public Disclosure Authorized A Country Perspective

Public Disclosure Authorized Jaime de Melo Arvind Panagariya and Dani Rodrik Public Disclosure Authorized

Regional integrationis on the rise again - but from different startingpoints and with differerntobjectives than in the past.

PolicyReaomdWodcing Pap dianatzthefndingsofwokinp awgresscncd .getheexchdngeofidesmongB staffaa ai oteheirmacdindevopncniLuespaps,&tibutewdbytc RusrchAdvis yStaff.cuaythenaresoftheauthorflact odytbavrvisndsuldbeLwddctaccordin0y.lefinduiginti, ndc

I TradePolicy

WPS 1094

This paper- a product of the Tradc Policy Division, Country Economics Department - is part of a larger cffort in the department to improve our understanding of the economics of regional integration. The paper was revised to reflect comments reccived at the Confcrence on New Dimensions in Regional Integration, (funded by the Bank's Research Suppon Budget underRPO 677-12).Copies of this paperare available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Plcase contact Dawn Ballantyne, room N 10-029, extension 37947 (February 1993, 47 pages).

Regional integration is on the risc again, despite arbitrage can Icad to improved economic its apparent failure among developing countrics outcomes by making decision-making less in the past. sensitive to economically harmful factional interests - especially when regional institutions De Melo, Panagariya, and Rodrik survey the are designed properly. ambig-ous economics of customs unions, emphasizing that the traditional dichotomy An empirical evaluation of existing schemes betwecn "trade creation" and "trade diversion" is produces no evidenec that membership in not particularly helpful for policy. In a world integration schemes has any effect on growth. with trade restrictions, regional integration presents certain advantages, including enhanced Finally, the authors note that recent attempts bargaining power and market access. at regional integration have differcnt starting points and objectives than past efforts - so The authors point out that integration history is a poor guide to the future of regional enforces arbitrage in institutions as well as in integration. markets for goods and factors. This kind of

Thc PolicyRcsearch Working Paper Scricsdisscminates the findings of workunder way in thcBank. An objectiveof the scries is to get thcse findingsout quickly, even if prescntationsare less than fully polished.The findings,interprctations, and | I conclusionsin these p '' not necessarilyrepresent official Bank policy.

Producedby thePolicy Research Disscmination Centc: The New Regionalism: A Country Perspective

by

Jaime de Melo World Bank, Universityof Geneva and CEPR

Arvind Panagariya World Bank and Univarsityof Maryland

Dani Rodrik Harvard University and CEPR

The authors are grateful to SumanaDhar, Claudio Montenegro, Francis Ng for superb assistance, and to conferenceparticipants for helpful cormnents. The New Regionalism: A Country Perspective

Table of Contents

1. Introduction ...... 1 2. Welfare Economics of Free Trade Areas .2 2.1 The Basic Economics of FrAS .3 2.2 Designing A Welfare-Improving FTA in the Presence of QRS .6 2.3 The Compensation Issue .7 2.4 FrA vs. UTL .8 2.4.1 Partner Country Tariffs .9 2.4.2 Third Country Tariffs.10 2.4.3 The Import Substitution Objective.11 2.4.4 The Tariff Revenue Constraint .12 24.5 Economics of Scale and Product Differentiation.13 2.5 FrA vs. PTA .15 2.6 FTA vs. CU.16 2.7 Strategic Advantage of an FrA .18 2.8 Inplications for Efficiency.19 3. Institutional Dimensions of RI.21 3.1 An Exploratory Model .. 23 3.2 Modeling Regional Intergration .25 3.3 Economic Consequences of RI.28 3.4 Political Consequences of RI.30

4. Growth Effects of RI Schemes.32 5. Conclusions.35

Endnotes . .38 References .. 41 1. Introduction

Three decades ago, under the impetusof European arrangements, the developingworld

launchedthe first wave of regional integration (RI). Free trade areas and customs unions

mushroomedin Latin America and Africa. Unfortunately,expectations of economicdevelopment

through regional integration were not realized and two decades later virtually all regional

arrangementsamong developingcountries were judged as failures.

By the early eighties, multilateraltariff cutting by developedcountries and unilateraltrade

liberalizationby developingcountries had substantiallyweakened the case for regional arrangements.

Yet, paradoxicallyit is then that a second round of regionalismgot under way. More arrangements

(eight) have been signed during the eightiesthan during the sixties, and still more (half-a-dozenor so) are under consideration. The GATr process is running out of steam and many countries are turning back to the bilateral alternative.

From the viewpoint of developingcountries, the current regionalismdiffer_ .,.om the regionalismof the sixties in two important respects. First, the regionalismof the sixties represented an extension of the import-substitution-industrializationstrategy from the national to the regional level and was therefore inward-looking. The current regionalism is by contrast taking place in an environmentof outward-orientedpolicies. Second, in the sixties, developingcountries pursued regionalintegration exclusivelywith other developingcountries. Today these countries, especially those in Latin America, have their eyes on integrationwith large, developedcountries.

In this paper, we review and extend the theory of regional integration and evaluate empiricallyits contributionto growth. Our objective is to assess the benefits of regional integration from the viewpointof participatingcountries rather than the world as a whole. In particular, we do not focus on the systemic implicationsof regional integrationemphasized in the precedingpapers. A central issue we address is whether the regional approach can accomplishobjectives that cannot be 2 accomplishedvia unilateral trade liberalization. We also study the role of economic institutionsin the process of regional integration. In analyzingthese issues, we draw a sharp distinctionbetween the nature of regional integrationtoday and that in the sixties.

In Section 2, we introducethe conventionalwelfare economicsof regional integration via freer trade among partner countries. We concludethe section by alluding to motivationsbehind current integration efforts which go far beyond trade integration. This theme is developed in greater detail in Section 3 where it is formally recognizedthat integrationenforces a certain degree of arbitrage among national institutions. In Section4, we provide an empirical evaluationof past integrationschemes. Finally, in Section 5, we turn to forward-lookinglessons for the 'new" approachesto regionalism.

2. Welfare Economics of FreeTrade Areas

To avoid confusion, we begin by definingthe terms preferentialtrading arrangement (PTA), free trade area (PTA), customsunion (CU), and unilateraltrade liberalization(UTL) precisely. A

PTA refers to an arrangement under which partner countries impose lower tariffs on imports from each other than on imports from the outside world. An FTA involves zero tariffs on trade among partner countries but positive tariff on imports from outside countries. Both PTA and FTA allow for different tariffs by partner countrieson importsoL similar goods from the outside world.' A CU is an FTA with a common external tariff by partner countries. Finally, UTL is defined as a non- discriminatoryreduction in trade barriers.

The literature on regional integradonis full of 'anything may happen' type of results. What we present below is what we regard as helpful insights from the literature.2 Unless otherwise noted, we assume throughout that the partner countries are small relative to the rest of the world. 3 In Section 2.1, we present the basic Vinerian analysis and derive conditionsunder which an

FTA is likely to be welfare improving. In Section 2.2, we consider the Kemp-Wanproblem of designinga welfare improvingFrA and apply it to the analysis of FTAs in the presence of qubotas.

In Section 2.3, we address the problem of compensationamong union members. In Sections2.4-2.6, we compare FTAs successivelyto UTL, PTAs and CUs. In comparingFTAs to unilateral trade liberalization,we discuss, inter alia, the roles of the import-substitutionobjective, tariff-revenue constraintand economiesof scale. In comparingFTAs to PTAs, we show that when chosen correctly the latter are superior, and then proceed to explainwhy the GATT approach of forbidding PTAs is, nevertheless,sensible. In comparingFTAs to CUs, we pay attention to the rules-of-originissue and political-economyimplications of the two regimes. In Section 2.7, wierelax the "small-union" assumptionand analyze the strategic advantagesof an FTA. In Section 2.8, we assess explicitlythe relevance of regional integrationbetween developingand developedcountries. Finally, in Section

2.9, we summarizethe main conclusionswhich follow from the review.

2.1 The Basic Econonics of FTAs

Can an FrA be welfare improving? Yes but not always. This is the central point made by

Viner in chapter 4 of his (1950) classic work, The CustomsUnions Issue. Viner introducedthe key conceptsof trade creation and trade diversion and concludedthat a trade-creating customsunion is welfare improvingwhile a trade diverting customsunion is welfare worsening.3

In the following, we introduce the conventionalanalysis of an PTA formally with the help of a two-goods,three-country model. The goods are denoted 1 and 2 and countries A, B and C. A and

B are potential partners in an FTA and C represents the outside world. In this setting, there are two possibletrade patterns: A and B import the same good or they import different goods. 4 In the case when A and B import the same good, they will import it from C and there will be

no trade between them in the initial equilibrium. Moreover, if the formation of an FTA leaves the

tariff on C unchanged, the initial equilibriumwill continueto obtain; the FTA will be vacuous. This

situation may well describe the reality of Frme regional integration schemes in developingcountries,

particularly Africa. In many of these schemes, th6 partner countries had very similar patterns of

trade and integration attempts had a very limited impact on trade patterns (see Foroutan, this volume).

In the more substantivecase when the partner countries import different goods, assume that

country A imports good I and country B exports it. For now, we concentrateon tariffs levied by

country A only. Later we introduceexplicitly tariffs in both B and C. In Figure 1, MAMAand

EBE,Bare the general-equilibriumimport-demand curve of A and export-supplycurve of B, respectively. The horizontal line pcpc is the relative price at which C is willing to buy and sell good

1 in the world market.

Autarky prices in A and B are given by the respective heights of their curves at their points of origin. Under free trade, the gains from trade are representedby the area under the import demand curve and above the world price for A, and that above the export supply curve and below the world price for B. Given constant costs and free trade in C, it neither gains nor loses from trade.

Assume that initiallyA levies a nondiscriminatorytariff at rate t on imports from B and C.

A's import demand curve as perceived by B and C is now given by MIAMI' where the latter lies below MiM by the amount of tariff per unit. The border price facing A is pc and total imports are RS (=DL). Of these, RH come from B and HS from C. Domestic price in A is given by pA 5 which is the height of the importdemand curve as perceived by A's residents. The gains from trade

in A are given by area mAID plus tariff revenu. DRSL and those in B by area RHEr.

Now introduce an FTA between A and B. Imports from B are no longer subject to a tariff.

As drawn, there is a "sudden death" of imports from C with all imports "diverted"to B. The union benefits B both because its terms of trade improve and its exports expand. The country's net gain equals area RHGV. The effect on A is ambiguousin general because, on the one hand, A's temis of trade deteriorate (or equivalentlyit loses all tariff revenue) while, on the other, the distortionbetween the domestic and (new) border price is eliminated. The country gains or loses as area LGW is larger or smaller than the rectangle RSWV. As drawn, the country loses. The effect on joint welfare of A and B is also ambiguous. They benefitjointly if the horizontally shaded area, LGU, is larger than the vertically shaded area, HSU, but loses otherwise. As drawn, the union as a whole benefits from the

FTA.4 That is to say, B's gains exceedthe losses of A and it can compensatethe latter.5

Figure 1 has been drawn in such a way that an FTA between A and B eliminatesthe imports from C entirely. This is done to highlightthe point that an FrA will generally generate both positive and negative effects and that the effect on the union's welfare is likely to be ambiguous.6 If we draw

B's export supply curve so that it crosses A's solid curve to the left of point L, the FTA does not eliminateimports from C. In this case, the internal price facing A is unchangedwhile its terms of trade with B deteriorate by the full amount of the tariff. B's share in imports rises but total imports remain unchanged. A's welfare declinesbecause its terms of trade with B worsen without any improvementin efficiency; B's welfare rises because its terms of trade improve and exports expand; and the union's welfare declinesbecause over some range imports coming from B cost more than C's price and there is no gain in efficiency in A.

What factors make the gains from an FTA larger or losses smaller? If we are willing to restrict ourselves to the case depicted in Figure 1 where both positive and negative effects are present, a number of points can be made.7 First, the higher the initial tariff in a given sector, the larger the

favorableeffect (area LUG) and smaller the unfavorableeffect (USH) of the FTA. Second, the lower

the post-FTA tariff on extra-unioncountries the less likely that the lower priced goods of the latter

will be displaced. Third, the higher the tariffs in the outside world on the partner, the larger will be

the gain or smaller the loss. In terms of Figure 1, the higher the tariff in C, the higher will be pc

facing A and B anthde smaller will be the area HSU. Fourth, the greater the complementaritya

import demandsof A and B, the larger the gains from FTA. In terms of Figure 1, the farther apart

are the import demand and export suipplycurves of A and B, the larger will be the gains (wea LGW)

and smaller the losses (area HSU).

On this last issue, it is worth noting that at low levels of income, complementarityis more likely to arise between countries that are different in terms of factor endowmentsand perhaps, therefore, countries with differentper-capita incomes. As we will discuss shortly (Section 2.3), this is precisely the conditionunder which the compensationissue is critical and implementation difficultiesmost serious. Complementaritiesamong high income countries are more likely to arise among countries with similar per-capita incomes. This is because the bulk of high-incomecountries' trade is of intra-industrynature which correlates negatively with differences in per-capita incomesof partner countries.

2.2 Designing A Welfare-Improving FTA In the Presence oi QRs

Can a welfare improvingFTA always be designed? Yes. This is the essence of the Kemp and Wan (1976) analysis. These authors demonstratethat under very standard assumptions,A and B can adopt a set of common external tariffs which improve their welfare without hurting the outside world. The essential idea is that the commonset of tariffs may be chosen in such a way that the 7 external terms of trade and hence the quantitiestraded with the outside world are unchanged while internal trade is rearranged to maximizethe gains from it.

The Kemp-Wanresult has important implicationsfor the analysis of FTAs in the presence of quantitativerestrictions (QRs). These implicationshave not been explored in the literature so far and will be considered here briefly. In Figure 2, we look at the two-good model. Suppose that initially

A restricts its total imports to G H by a global quota. Assumingthat quota licenses are auctioned competitivelyto domestic residents, quantity GL will be imported from B and LH from C. If we now introducean FTA such that C is subject to a quota at its original level of imports (LH) while imports from B are freed of any restrictions,total imports will expand. Subtractingquantity LH horizontallyfrom MIAMAeverywhere, we obtain MAMA'as the demand curve facing B. This yields SF quantity of imports from B and FN (=LH) from C. Once again, as in Figure 1, we have a positiveeffect (horizontallyshaded area) on A due to reduced gap between the domestic and border prices and a negative effect (verticallyshaded area) due to a deteriorationin the terms of trade with

B. The net welfare effect on A is ambiguousin principle although as drawn it loses.

Country B necessarilygains in this case. Tbis gain, representedby area SFRP', is larger than the vertically shaded loss of country A. Therefi ,e, B can compensateA, should it lose from the

FTA on balance. The FTA leads to unambiguousbenefits for the union as a whole. In the spirit of the Kemp-Wanresult, the FTA has not .:duced trade with the outside world and has expandedtrade between the partners. The outside world's welfare is unaffectedso tht the FTA improves the world welfare as well.

2.3 The Compensation Issue

In the preceding sub-sections,we have consideredregional integration from the viewpoint of the union as a whole. We implicitlyrelied on lump sum redistributionsbetween partners to ensure 8 that neither partner loses. In practice, compensationmechanisms are difficult to implement,

particularly in developingcountries. Because a detailed discussionof the issue is contained in the

paper by Foroutan (forthcoming),we confine ourselves here to brief remarks.

Partner countries in integrationschemes in developingcountries often have very diverse levels

of per capita incomes. If integrationleads to a migration of industriesfrom the poor to the relatively

more advancedeconomies, compensation is essential. In some cases, special funds are crevted to help

the industrial developmentof the poorest members. The criteria for contributionsto and

disbursementsfrom these funds often becomecontentious, leading to very complicatedformulas that generally defy economiclogic.

In the case of CEAO and UDEAC in Africa, unions have provided for preferentialduties on

industrialproducts based on the protection needs of the poorer of the members. This has resulted in different tariff rates for the import of the same product depending on the source. The duty is lower on imports comingfrom the least developedmembers and higher on others.

2.4 FTA vs. UTL

We now consider what is perhaps the strongest criticismof the regional approach from the viewpointof small economies: An FTA is usually dominatedby unilateraltrade liberalization.

To begin with, observe that in Figure 1 country A's welfare is maximizedby liberalizingits trade unilaterally. That is to say, the FIA with B is welfare inferior to unilateral free trade.

Moreover, the same conditionswhich make FTA desirable (e.g., high initial tariff) make unilaterat trade liberalizationeven more desirable. Thus, there is nothing which B can offer A that the latter cannot get on its own. Why should A then bother to enter a bilateral arrangementwith B?

A number of complicationsmust be introducedto answer this question. Before we do so, however, it is usefil to introducea dramatic example. Suppose the developedand newly 9 industrializedcountries are divided into three blocs: North America, Western Europe and East Asia.

Also suppose that each one of these blocs allows free internal trade but imposes high duties or

voluntary export restraints on extra-bloc imports. Under these circumstances,will developing

countries in Aftica, Asia and Latin America benefit more from unilateral trade liberalizationor from

joining one of the blocs? Common sense dictatesthat the latter option is likely to be welfare superior

even though it involves adoptingand maintainingthe bloc's barriers to trade with extra-bloc

countries. In sections 2.4.1 and 2.4.2, we develop the argumentunderlying this example.

2.4.1 Partner Country Tariffs. A natural starting point is the introductionof tariffs in B.'

Figure ; assumes that B is a free trader even before the FTA is formed. This means that B has

nothing to offer A as a part of the FTA agreement. In practice, B will have tariffs initially and an

FTA will buy A a tariff free access to B's markets. In terms of Figure 1, the deteriorationin A's

terms of trade which accompaniedthe FTA need not take place. Indeed, B may end up offering it better terms of trade than C, making FTA a better option than unilateral liberalization. It turns out, however, that this cannot happen unless B chooses a suboptimaloption for itself.

In Figure 3, solid curves show A's general-equilibriumdemand for imports of good 1 and B's general equilibriumsupply of exports of the same good under free trade. The horizontal line gives

C's supplyprice which A and B are too small to influence. Both A and B impose tariffs in the initial equilibrium. The tariff in A causes its import demand as perceived by B and C to shift down to the dashed curve. The tariff in B causes its supply-of-exportcurve as perceivedby A and C to shift up as shown by the upward sloped dashed curve.10 In the presence of nondiscriminatorytariffs, A and

B trade along dashed curves at the price eifered by C. A imports ML and B exportsRL. It is not necessaryfor A and B to trade but if they do, B exports up to ML to A and MR to C. Domestic prices in the two countries are given by their respective solid curves. For each country, the gains from trade are measured by the area enclosed by the world price, quantitytraded, and its solid curve. 10 If A now liberalizes unilaterally, its imports will rise to LN and its welfare will improve.

The critical question is can A do better by forming an FTA with B. Suppose A and B eliminate tariffs between themselvesand impose prohibitivetariffs on C. This will yield G as the equilibrium with A benefitting more than from unilateraltrade liberalization.

Does this suggest that the FTA in the present case is superior to unilateral free trade? The answer is in the negative. For we have noi said anythingyet about B's welfare. As is easy to see from Figure 3, compared to point G, country B can improve its welfare by liberalizing unilaterally and moving to point S. The FTA can be superior only if A could compensateB to ensure the latter the income level at S and still come out ahead relative to unilateral liberalization. This is impossible, however, since the amount by which A's incomeat G exceeds its income at N is less than the required compensationby area NGS. Looked at differently, combinedgains from trade of A and B under unilateralliberalization are larger than those under an FTA by area NGS.

The conclusionfrom this discussionis that partner country tariffs do not negate the overall superiority of unilateral trade liberalizationover regional integration. Only if one of the countries - country B in Figure 3 - is willingto accept a suboptimalposition for some exogenousreasons (e.g., political hegemony), can the other country benefit more from an FTA than from unilateral liberalization.

2.4.2 Tbird Country Tariffs. Anotherpossible explanationfor a preference for an FTA over unilateraltrade liberalizationis the presence of trade restrictions and/or transport costs in the outside world, country C. Thus, suppose that there are tariffs initially in not just A and B but also in C.

This means that if C is a buyer of good 1 in the world market, it will offer a price lower than its internal price. Similarly, if C is a seller of good 1 in the world market, it will charge a price higher than its internal price. 11

In Figure 3, suppose now that C's selling price is pc, its buying price is pc', and its internal

price is somewherein the middle. This means that B no longer has the option of trading at R.

Indeed, the terms offered to it by C, pc', are worse than those offered by A. Likewise for A, the

termasof trade offered by B are better than those offered by C. The two countries will trade with

each other at H. Now if A and B eliminatetariffs on each other, they can trade at G and both do

better than if only one of them was to liberalize trade unilaterally. The FTA turns out to be superior to unilateraltrade liberalization.11

An important point to note is that in this example, there is no difference between an FTA and

complete free trade by A and B. High tariffs in C make that country irrelevant for optimal trade

policies in A and B. In order to reach G, A and B do not need any external tariffs. In this sense, the

only role of an FTA is to solve the prisoners' dilemma for the two countries and bring them to the negotiatingtable.

A useful applicationof the exampleprovided in this subsectionis that if the world gets divided into inward-lookingblocs, unilateral trade liberalizationwill become a less attractiveoption for the countries outside the bloc than it is today. The countries will then be better off either seeking access to one of the blocs and adoptingits trade policy or engaging in regional integrationso as to promote freer trade among themselves."2 Of course, the current world being quite far from consistingof closed blocs, the example does not justify a preference for regional integrationover unilateraltrade liberalization.

2.4.3 The Import SubstitutionObiective. The most commonjustification for pursuing regional integration instead of unilateral liberalizationcomes in the form of an implicit or explicit noneconomicobjective. In the early developmentliterature, industrializationvia import substitution was considered an eminentlyrespectable objective. This, either on grounds of infant industry, 12 "trainingground argument" or other reasons, meant that the optiuuof unilateral free trade was simply

not available. Under such circumstances,an FTA which, in the spirit of Kemp and Wan, keeps the

imports of industrial goods from country C fixed and exploitsthe gains from specializationbetween partner countries is beneficial. 1t

During the sixties, the import substitutionobjective played a key role in the proliferation of regional integration schemes in the developingworld. But the gains from specializationexpected of various schemes were not realized for reasons we need not go into here. Suffice it to note that today the policy environmenthas changed dramaticallyand import substitutionas a policy objective has fallen out of favor with policy makers except perhaps in the early stages of development. This means that an FTA as an alternativeto unilateraltrade liberalizationis harder to defend on grounds of import substitution.

2.4.4 The Tariff Revenue Constraint. Another objective which may preclude free trade as a viable option is tariff-revenueconstraint. In many developingcountries, domestic-taxmachinery is limited and they must rely on trade taxes as the major source of r3venue. In this situation, the option of unilateral trade liberalizationis simply not availableand we can ask whether regional integration may be a superior option than nondiscriminatorytariffs.

For concreteness,assume that A and B choose their initial revenue-constrainedtariffs optimally. With some qualifications,this implies relativelyhigh tariffs on goods with low elasticities of demand. If A and B now form an FTA, they will have to restrict their tariff structure so as to impose no tariffs on each other's goods. Under standard assumptions,including fixed world prices, this restriction will lead to a lower level of welfare. The same forces which make unilateral free trade superior to an FTA under standard assumptionsmake nondiscriminatorytariffs superior to preferential tariffs in the presence of a revenue constraint. 13 2.4.5 Economiesof Scale and Product Differentiation. During the first wave of regional

integration arrangements,many economistsrelied on the notion that an FTA can serve as vehicles for

the exploitationof economiesof scale. Yet, broadly speaking, scale economiesby themselve do not

make regional integration superior to unilateral trade liberalization.14 Intuitively, if the minimum

cost of production of a good along the long-runaverage cost curve is below the world price, both potential partners in the FTA should expand production until the marginal cost is less than or equal to the world price. They should a.en consumedomestically as much as is demanded at the world price and export the residual. Goods for which minimumcost is above the world price should not be produced. Unilateral free trade will generally ensure this outcome.

If the case for regional integrationover unilateral trade liberalizationalready exists due to considerationscited in sections 2.4.2 or 2.4.3, scale economiesgenerally make it much stronger.

This is perhaps the reason why economiesof scale are often a part of the case for an FTA. The point can be illustrated with the help of two examplesof which we develop the first one in detail and refer the reader to Corden (1972) for the second one: (i) an import substitutionobjective involvinga target level of industrial output as in Section 2.4.2; and (ii) tariffs and transportationcosts on exports to the outside world which limit access to the latter's market as in Section 2.4.3.

Supposethat both A and B have a target level of aggregate import-competingindustrial output. As noted before, starting from a nondiscriminatorytariff, the two countries can benefit from an FTA under such circumstanceseven under constant returns. If scale economiesare present, however, the gains from specializationare likely to be larger. For specializationin the presence of scale economiesyields gains in the form of larger rectangular areas rather than conventional triangular areas. Thus, the case for an FrA is stronger.

The above analysis assumes (implicitdy)that products are homogeneousand the scale economiesare external to the firm.l" Supposenow that scale economiesare internal to the firm and 14 productsare differentiated. These assumptionsdrive us automaticallyinto the world of imperfect

competition,where individualfirms are able to exercise market power. In view of market power on

the part of firms, potential partners in an FrA cannot be viewed as small in relation to the outside

world. Therefore, their optimal tariffs are positive and UTL is nonoptimaleven in the absence of the

import substitutionobjective.

Beyondthis point, the implicationsof imperfect competitionare rather complex. Depending

on the behavioralassumptions, a myriad of models can be constructed.16 We resist temptationand

restrict ourselvesto the monopolisticcompetition model. On balance, the results which emerge from

this model are intuitivelymore plausiblethan those arising from oligopolymodels." 7

Assume that potential partners consumedifferentiated goods for which markets are

monopolisticallycompetitive and that consumersappreciate variety. Startingfrom an initial protected

equilibrium,formation of an FTA can yield two additionaltypes of gains not available in the conventionalmodels. First, after the FTA is formed, each partner will have tariff-free access to the varietiesproduced by the other. To the extent that consumersprefer more variety to less, welfare must rise. Second, as the market for each variety is likely to be larger in the post-FTA equilibrium, its scale of operation will expand. Given decreasing costs, this will yield further gains.

Against these gains, we must weigh the losses which are likely to accrue from increased distortionbetween varieties imported from the partner country and those imported from the outside world. Preferential tariff reductionwill reduce imports from the outside world, which is harmfil for members of the FTA. Although, in general, the net result is ambiguous,there is a strong presumptionthat due to decreasing costs welfare will rise. Indeed, if the members freeze their vectors of imports from country C, the FTA will be unambiguouslywelfare-improving.

Two important policy implicationsof this analysis may be noted. First, to the extent that intra-industrytrade is predominantly among rich countries, these gains are less relevant to FTAs 15 between poor countries. Second, contrary to the results based on models of inter-industrytrade,

similarityamong partner countries does not diminishthe case for FTAs when potential trade is of the intra-industryvariety.

2.5 FIA vs. PTA

We now show that preferential trading arrangementscan be superior to an FTA. That is to say, partial preferencescan yield a better outcomethan a 100 percent preference granted under FTAs.

The point is made most easily with the help of a three good model in which each of A, B, and C exports one good and imports two goods." Goods imported into each country from the other two - say, those into A from B and C - are imperfectsubstitutes. A and B are small with respect C so that their terms of trade are fixed. Given symmetry in the trade pattern, we need to consider the effects of a PIA on one partner only, say A. The effects on B are similar.

Denote the goods exported by A, B, and C by 1, 2, and 3, respectively.

Assume that initially, A imposesthe same ad valoremtariff on B and C. That is to say, the inidal marginal distortion across B and C is the same. Supposenow that the tariff on B is lowered by a small amount. This will raise imports of good 2 and, given substitutability,lower imports of good 3 and raise exports of good 1. Because exports rise, there is a net expansionof imports implyingthat imports of good 2 expand more than the contractionof imports of good 3 at world prices. The inidal distortionin the two imports being the same, the gain from increased imports of good 2 is larger than the loss due to reduced imports of good 3. The PTA is welfare improving.

As we lower the tariff on good 2 further, net imports will continue to expand. But the gain from the expans of imports of good 2 will be evaluatedat the lower tariff than the loss from the contractionof imports of good 3. Therefore, there is no guaranteethat welfare will continueto rise as we lower the tariff on good 2. Indeed, assumingmonotonicity, we can find a critical, second-best 16 optimal tariff at which the gains in good 2 from a small additionalreduction in tariff are exactly offset

by the losses in good 3. Any reductions beyondthis tariff rate will lower welfare. By obvious extension, an FTA which pushes the tariff on good 2 down to 0 will yield a lower welfare than the

PITAwith the second-bestoptimum tariff.

This conclusion clearly brings into question the wisdomof the GATT Article XXV which prohibits PTAs but permits FTAs and CUs. Viner provided at least two reasons why in practice a

100 percent preference may be superior to somethingless than that. First, there are likely to be some economiesin administrativecosts. "The burden on trade of a customs tariff,..." reasoned Viner,

"arises...also from the costs involved, for exporter and importer, in meeting the customs regulations, and the costs involved, for the tariff-levyinggovernment, in administeringthe customs machinery.

These costs are often, in fact, more important than the duties themselvesas hindrances to trade..."

A second and perhaps more important reason for the preference for an FTA over PTA is that even though superior PTAs exist, there is no guaranteethat those are the ones that will be picked. It is entirely possible, even plausible, that the PTAs that are actually picked are inferior to the FTA.

Once again, it is worthwhileto quote Viner,

CustomsUnion...involves a cross-the-boardremoval of the duties between the members of the union; since the removal is non-selectiveby its very nature, the beneficial preferencesare establishedalong with the injuriousones, the trade-creatingones along with the trade- diverting ones. preferential arrangements,on the other hand, can be, and usually are, selective, and it is possible, and in practice probable, that the preferences selected will be predominantlyof the trade-divertingor injurious kind. (Viner, 1950, pp. 50-51).

2.6 FTA vs. CU

An issue of great significancein the formationof regional integrationschemes is whether the countries should retain their own tariffs (FTA) or erect a common extemal tariff (CU) on extra-union partners. Economistsare divided on this issue. 17 There are three main arguments given in favor of FTAs. First, they do not result in an

increase in tariffs in member countries. By contrast, when a commonexternal tariff is imposed,

countries with lower initial tariffs are able to raise their tariffs. This is because Article XXIV of the

GATT requires that the external tariff be no higher than the average of the tariffs in the partner

countries prior to the formation of the union. Second, an FTA gives the member countries a greater

freedom to pursue their trade policy reforms. In a customs union, more reform-orientedcountries

may find their hands tied by the agreementto maintaina commonextern-al tariff. If lobbies are

powerful, the common external tariff may be set at a level high enough to protect producers in the

least efficient country in the union. Finally, due to enforcementdifficulties under an FTA, the union

finds itself importing goods through the border of the country with the lowest tariff. This puts

pressure on countries with higher tariff to bring their tariffs down to the level of the least protective

country. Thus, there are dynamicprocesses at work in favor of ever declining tariffs under an FTA.

In principle, at least some of these arguments can be turned on their heads. First, if a customsunion is formed, the least protective country can force the more protective countries to lower their tariffs to its level. This will make the likelihoodthat integration is welfare improvinghigher and result in a lower cost to the rest of the world. Second, in practice, FTAs are accompaniedby elaborate rules of origin and content requirementwhich become powerful instrumentsof protection.

In a customs union, such rules are ncit required. Finally, if protection lobbies are active, their effectivenessmay be diluted in a customsunion since the latter requires union-widelobbying, whereas under an FTA tariffs are responsiveto the lobbying at the national level. In the ultimate, the issue would seem to hinge on whether we can expect institutionalarbitrage across countries (in the sense of section 3) to lead to superior outcomes for all, in which case a CU would be preferable. If such arbitrage leads to worse outcomes, then it is better to leave each country with its own external tariffs.

Perhaps put differently, if the "center" country, when one exists, has a clean trade policy, a CU can 18 be more desirable, since we can expect that region-widetariff policies will be determined more by this country.

2.7 Strategic Advantages of an FIA

Up to this point, we assumedthat the union is small in relation to the outside world. We now allow these terms of trade to change and consider the strategic advantagesof an FTA for union members. Our discussionis in .,o parts. First, we summarize the effects resulting directly from the formation of an FTA. Second, we discuss the gains which may arise from increasedbargaining power of the union vis-a-vis the outside world.

In the two-good model of Figure 1, the terms of trade effects of an FTA are straightforward.

Suppose that C's supply price is not constantbut increases with quantity. At a constant price of C, formationof an FTA reduces the demand for imports from C. This leads to a decline in C's price.

Formationof an FTA improvesthe union's terms of trade vis-a-vis the outside world."9

The second type of gain from an FTA can result from increased bargainingpower of the union. The terms of trade of union members depend not merely on their external tariffs, but also on tariffs imposed on them by the outside world. To the extent that the level of these tariffs can be influencedthrough bargaining, an FTA which increasesthe joint bargainingpower of member countries can confer further gains on the latter. Interestingly,unlike the first type of gains, these gains need not come at the expense of the outside world. The union as well as the outside world may benefit from mutual tariff reduction.'

Viner provides several examplesfrom history where decisions to forge economic unionswere influenced,in part, by expected gains from increasedbargaining power with respect to other countries. In the U.S., the Articles of Confederationhad left each state with its separate tariff. The argumentthat this put the United States at a disadvantagevis-a-vis Britain and other European 19 countries in matters of commercialpolicy was instrumentalin mobilizingpublic opinion in favor of a federal union with a centralizedtariff policy. The recent movementfor a "United States of Europe' sprang in large part from the feeling that a European economicunion was necessary to deal with a large America.

A case can be made that the existence of the EC has been instrumentalin allowing Western

European countries to cut out better deals with the U.S. in bargaining tariff reductions than would have been possible if they had dealt individuallywith the U.S. More recently, the restraint shown by the U.S. towards the EC in using Super 301 threats is a direct result of the EC's ability to inflict injury on the U.S. through retaliation. By contrast, the U.S. has used this instrumentwith relative ease against individualcountries such as Japan, Brazil and India.

Bargainingpower of a union is a direct function of its economicsize relative to the outside countries with which it must negotiate. In this respect, formationof the EC did prove beneficialto its member countries. By contrast, economicunions in Africa and Latin America were much too small relative to their counterpartsin the developedworld (see papers by Foroutan, and Nogues and

Quintanilla). As a result, any expectationsof these countries regarding benefits from joint negotiationswith the U.S. and the EC were bound to result in a disappointment.

2.8 Implications for EMdency

The above discussions give some clues about the likely effects of RI arrangements. First, countries with high barriers to trade (e.g., India) are likely to benefit most from unilateral trade liberalization. The gains from such liberalizationare likely to be much bigger than those from

FTAs. Only if the world is divided into closed trading blocs is regional integration a superior option.

Second, in the absence of closed trading blocs, the case for FTAs must be based on an exogenous objective such as import-substitutionindustrialization. Third, economiesof scale by themselvesdo 20 not provide a reason for justifyingregional integration over unilateraltrade liberalization. If there are

other reasons for a preference for regional integration, however, economiesof scale can reinforce

them. Fourth, regional integrationamong low-income,developing countries is unlikely to yield major

gains. This is because either these countries importvery similar goods from developed counitriesor

differencesin income levels among them lead to very diverse distributionof gains. In the former

case, gains from specializationbetween partner countries are limitedwhile in the latter case

compensationschemes which emerge are highly distortionary. Fith, similarity in per-capita incomes reinforces the gains from regional integration among developedcountries. Sixth, a priori, it is not clear whether an FTA is superior to CU. Much depends on how these are implemented. Finally, the larger an FIA, the greater the strategic advantages. For instance, the EC has been generally insulatedfrom the Super 301 threats of the U.S. while countries such as Japan, India and Brazil had to face these threats.

As a prelude to the next section, nothing has been said explicitlyso far about North-South integration, which has been gaining popularityin recent years, e.g., Mexico and the U.S. Countries engaged in this type of integrationhave relatively free trade regimes (the average Mexican tariff is 9 percent). The gains for a developingcountry from integrationwith a rich country go beyondthe trade efficiency gains discussedhere. First, should the world get divided into inward looking blocs, such integrationguarantees future access to a large market.21 With this large market, the competitivepressure to ensure efficiency and availabilityof new technology is also guaranteed.

Second, such integration involvinginternational obligations ensures that the country's reforms will not be reversed. This is critical for the credibilityof the country's policies. Finally, as a part of the integration effort, the country may be able to import pro-growth economic institutions(e.g., more liberal labor markets, stable macroeconomicpolicies, etc.) of the developed country.

We now proceed to develop this last point in greater detail. 21 3. Institutional Dimensions of RI

The implicationsof RI go beyond trade in goods, services, and factors. Almost by

definition, any regional arrangementworth its name entails the impositionof some common rules of

conduct on the countries entering the arrangementand a set of reciprocal commitmentsand

obligations. Regional integrationthereby enforces a certain degree of arbitrage among national

institutions,just as it brings about arbitrage in markets for goods and services. The importanceof this political dimensionof RI may well exceed that of the more direct implicationshaving to do with trade flows.

That the effects of RI can extend beyond trade has long been recognized. More than two decades ago, Hirschman argued that surrendering autonomyto supra-nationalauthorities had both costs and benefits for governments:

[Clommonmarkets would not only provide preferential treatmentfor the industrialistsof the participatingcountries; for these mutual arrangementsto be durable, monetary and foreign exchangepolicies would have to become more uniform and stable than they have been; and such a development would be even more importantthan the customs preferencesthemselves in promoting exports from the common market countries, not only to each other, but also to third countries. It is, however, precisely the prospect of less freedom of movement in monetaryand foreign exchangepolicies which makes national governmentsso skittish about entering effective common market commitments.(Hirschman, 1971, 122.)

The issues go well beyond monetary and foreign exchangepolicies, and skittishnessabout losing autonomy is not confined to them alone. RI can also serve as a conduit for industrialpolicies, environmentalpreferences, social and welfare policies, and so on.

If the cost of integrationis reduced autonomy, the benefit (as Hirschman suggests) may be superior economicoutcomes. Much of the recent trend towards regional integrationcan be understoodas reflecting the desire of certain countries to 'borrow" or "import" desirable institutions from their neighbors. A central aim of the exchange-ratemechanism of the European Community was to enhance the credibilityof anti-inflationarypolicies by effectively surrendering monetary 22 autonomyto the Bundesbank. By entering a quasi-irrevocablearrangement in the context of the

NAFTA, the Mexican governmentis as anxiousto cement its new, open trade policies as it is in

obtainingspecific trade privileges via the arrangement.22In both cases, regional integration is used

to committo desirable policies in a context where discretion is feared to produce sub-optimalresults.

The literature on rules versus discretion providesthe natural reference point for evaluating

such institutionalimplications (see Fischer, 1990, for a survey). However, we shouldbe clear that

membershipin an RI scheme does not purchase commitmentin any direct way. The arrangements

that occur in practice rarely involvethe complete subordinationof a member's preferences to

exogenousrules or to the preferencesof another member. We will highlighthere three other

channelsthrough which regional institutions can alter economicoutcomes.

(a) The preference-dilutioneffect. Irrespective of the institutionalsetup, a regional arrangement implies a larger political communityand hence a smaller role in determiningpolicy for politically-importantgroups in each of the countries. This renders decision-makingless responsiveto factional interests, and may thereby enhanceefficiency.

(b) Thepreference-asymmetry effect. Unless members have identicaleconomic structures and preferences,policy-making at the regional level will have to compromiseover the perceived needs of different countries. Somewhatparadoxically, the compromisesolution may present a more efficient outcome for some members than could be ootained in the absence of integration. For others, the compromisesolution may be worse.

(c) The instiutional-designeffect. Within establishednation-states, policy makers have to live with the inherited institutional setting. But when a regional institutionis set up from scratch, it may be possible to optimize in the choice of certain institutionaldimensions in a way that would not normally be possible in the domesticcontext. This greater flexibilityin institutionaldesign may enhanceefficiency in all members. 23 3.1 An Exploratory Model

To make these arguments more concrete and to draw some implications,we now turn to a

simple model. As in the rules-versus-discretionliterature, our focus is on the interactionbetween

governmentsand their private sectors, and the changes in this interaction that would follow from

integration. We develop the case of a single country first, and then turn to the case where two countries become integrated.

We suppose that governmentbehavior can be summarizedby a quadratic-loss objective

function which allows for both an economicmotive and a non-economic("political") motive:

2 a = [(g_g)- + y (e g9v (1)

The government's choice variable is denoted by g. It representsgovernment interventionon some

relevant policy dimension(e.g., trade protection, industrialsubsidies, aggregate demand-management policies, etc.). The optimal level of g in a strictly economicsense is given by j. Hence, the first term in the above expression captures a purely economicmotive in govermnentdecision-making. In addition, we suppose that the government comes under pressure to intervene on accountof political motivesalso, representedhere by the second term in (1). Politically-powerfulgroups in the private sector have a most preferred level of interventiongiven by 0. The second term captures the idea that the governmentpays a penalty whenever g diverges from e. y is the relative weight placed by the governmenton the political motive. We attach no normative importanceto this objective function; it is simply meant to characterize governmentbehavior in a transparent fashion.

The level of 0 is endogenous,and is selected by a lobby representing politically-active groups in society. These groups derive private benefits from govermnentintervention, but they also bear a cost whenever g diverges from the economicallymost efficient level. This trade-off is representedby a utility function of the followingsort: 24

U = ag -(6- _)2 (2)

The first term here representsthe private benefits of governmentintervention, while the second term

standsfor its economy-widecosts. These costs are r3ositivewhenever the private-sector lobby

expresses a preference for a g that is different from g, and they are assumed to be partly internalized

by the lobbying group.

The game proceeds as follows. In the first stage, the private-sector lobby announces e, its

most preferred level of govermnentintervention, taking into accounthow the governmentwill respond to this announcement. In the second stage, the governmentdetermines its optimal level of g, taking e as given. As we shall see, in this sequence the governmentsuffers a loss in welfare from being unable to precommit to a specifiedlevel of g in advance of the political process.

We solve the second stage of the game first. Maximizing(1) with respect to g yields:

g=(1 +y)-'(g+yO) mg(e) (3)

Tlhisis the government's best-responsefunction with respect to e. In the first stage of the game, the lobby in turn maximizes (2) subject to (3). The solution is:

6=j+c y/2(l +y) (4) which is larger than g as long as a > e. Substitutingthis back in (3), we get the equilibriumlevel of governmentintervention:

2 g =g + (a/2) [y/(l + y)] (5)

Hence g exceeds g as long as a and -y ae both positive. Therefore, there is more interventionthan is economicallydesirable. We can also show that:

(e - g) =(ae/2) 1Y/(l + y)J > O, (6) 25 which implies that the governmentchooses to moderate the lobby's demands. For future reference,

we plug (5)-(6) in (1) to express the value taken by the government's objective function in the non-

integrated equilibrium:

ON = -(a2/4) [y/(1 + y)] 3 (7)

By being responsiveto political pressure, the governmentsuffers a cost. To see that this

cost originates from the specifictiming of moves adopted here, suppose that we reverse the order of

moves. Now the governmentcan precommitto a specific level of g. Since the lobby moves second,

it must take g as given. In view of (2), its optimal O is then equal to j. In turn, the government

can maximize (1) by setting g = j. Hence, when the governmentcan precommit, its objective

function attains its bliss point (of zero). We can interpret this framework also in terms of time-

inconsistency:the government's optimal ex-ante policy diverges from its optimal ex-post policy. We note that the private-sectorlobby is the beneficiaryof this time-inconsistency,as it attains a higher level of utility when the governmentcannot precommit than when it can.

3.2 Modeling Regional Integration

We now turn to the case where two countries decide to integrate. We model this integration by requiring that the level of governmentintervention be the same in the two economies. We suppose that the two countries differ only with respect to the economicallydesirable level of intervention.

Hence, the foreign government's objective function is given by:

Go= -[Ri - g y (80 -g 2 yo, (8) while the private-sectorlobby in the foreign country maximizes:

U. = ago - (0 -_j (9)

Note that a and y are commonto the two countries. 26 How is decision-makingaccomplished at the regional level? It is natural to think that the

regional institutionwill maximizean objective function that has the same structure as the ones in the two countries:

2 2 = _- ,_ g) + y( - g) ], (10)

where the subscript 'r" refers to the regional institution. To save on notation, we do not subscript g,

as this is now commonto both countries - it may stand, for example, for the commonexternal tariff.

As this expression makes clear, there are basically two issues here: first, given that g and j

differ, what objective does the center see itself as fulfilling (i.e., what is j,)? Second, as the respective lobbying demands e and e may in principle differ also, how does the regional institution

"aggregate' these preferences (i.e., what is the relationshipbetween e and e, on the one hand, and

e, on the other)? We make what seems to be the neutral assumptionby equating gt and Or to the averages of the preferences in the two countries. That is,

2*=1i+g) (12)

e,-= 2(e+e@-) (12)

Note that having the regional institutionmaximize the sum of the two governments'objective functionswould yield equivalent results to the approach taken here."

Now we can solve for the outcomeof the game under regional integrationonce again by working backwards. The regionalinstitution's best-responsefunction is:

g = (1 + y)t + y 0) a g (0)- (13) 27 The national lobbies in turn maximizetheir objective functions, internalizingthis best-response function, but taking the actions of the other lobby as given. Using (12), this yields:

e=i+cay/4(1+y) and e= j ay/4(1+y) (14)

By comparing(14) with (4), we can see that regional integrationmoderates the national lobbies' demands for intervention. The reason is simple: each lobby now has a smaller impact on decision- making, as the central institutionhas to contend with not one but two groups clamoring for attention.

Since the marginal benefits of lobbyin- have gone down, the groups rationally choose to do less of it.24 This is the preference-dilutioneffect mentionedabove.

That larger political communitiesmay be less susceptibleto harmful factionalismis in fact an old insight. This was one of the main argumentsused in The Federalist Papers in support of a

"well-constructedUnion" of American states. As James Madison argued, individualfactions could well come to dominatethe political agenda in small political communities,but they would be unlikely to do so in a federal setup where they would have to contend with many more pressure groups:

The smaller the society, the fewer probably will be the distinct parties and interests composingit; the fewer the distinct parties and interests, the more frequendy will a majority be found of the same party; and the smaller the number of individualscomposing a majority, and the smaller the compass within which they are placed, the more easily will they concert and execute their plans of oppression. Extend the sphere, and you take in a greater variety of parties and interests; you make it less probable that a majc.ity of the whole will have a common motive to invade the rights of other citizens; or if such a common motive exists, it will be more difficult for all who feel it to discover their own strength and to act in unison with each other. (Hamiltoneltal., 1981 [1787J, 22.)

However, note that e and e also depend on g and jg, respectively, so that the overall pressure for interventionmay not be reduced for both countries simultaneously. Combining(12) and

(14),

Or =g+ a y/4(1 + y) = (gC + ) + a y/4 (1 + y) . (15) 2 28 On account of the second term, each of the two countries experiencesa reduction in lobbying (cf.

[4]). But the effect of the first term works differently for the two countries. This is the preference-

asymmetryeffect mentionedabove. If g > g, the home country gets an additionalreduction in

lobbyingfrom regional integration,while the foreign country gets an offsetting increase. Using (14),

we can now state the equilibriumlevel of interventionin the union:

=1 i +j) + (a/4) [y/(l+ y)]2 . (16) 2

Comparingthis with (5), we can see that the direction of change in g after integration dependson the

relative magnitudesof the preference-dilutionand preference-asymmetryeffects. The first effect pulls g down in both countries, while the second effect reduces g in one country but increases it in the other.

Now we are ready to evaluate the consequencesof integrationfor each of the two countries.

We will do this from two differentperspectives, and ask in turn: (i) does integration lead to more efficient outcomes from a strictly economicstandpoint? and (ii) does integrationmake each of the governmentsbetter off in terms of their own objective functions, includingtheir political motives?

3.3 Economic Consequences of RI

The economicevaluation hinges on whether integrationhelps close the gap between the economicallydesirable levels of governmentintervention and the actual levels chosen in equilibrium.

2 That is, we need to check whether (i- g) and (* - g o are smaller under integration. Let us denote the economiccomponent of the government'sobjective function by G3and use the subscripts

"N" and "I" to refer to the non-integratedand integrated cases respectively. Then:

ON = N I-g)IN(sf/4) [y/(l + y)]4, (17) 29

C} = _i-g)2ll=-(+y)]2)2 1(i-g)-(a/4)[y(l( (18) 2

t~~~( ~~~g2 = - )-(a4 l(l + y)]2)2. (19) 2

(Asterisksdenote the foreign country as before.)

Consider first the case where a - = o, that is when the two countries are identical. By inspection, we can see that in that case C.= GN< Gi = X . This is due entirely to the preference- dilutioneffect, and represents the beneficialeconomic consequenceof a larger political union. This effect operates for both countries. When (i - le) * o, on the other hand, the preference-asymmetry effect comes into play. For "small" differences,this effect benefits the country with the larger level of economicallydesirable intervention. The reason is as follows. In the absence of integration, govermnentintervention is over-providedin both countries. Under integration, the preference-asymmetryeffects works to reduce g in the country with the higher level of desirable interventions,and to increase it in the other country. Given the initialdistortion, this benefits the first country and hurts the second. For a "large" divergencebetween j and j, however, kQU countries are made worse off by integration.

Finally, note an importantimplication. As the number of countries entering an arrangement increases, the conditionthat each must be made better off becomes more and more stringent, as long as the countries are not identical. Hence the relatively small number of participants is one significant advantageof regionalismover multilateralism.21

We have assumedso far that the regional institutionselects as its target for interventionthe simple average of the two countries' targets [i.e., g, = I i + g . As discussed above, this creates a 2 30 positive externalityfor the high-interventioncountry and a negative externality for the low- interventioncountry. Now suppose that the regional institution can be designed in a manner that would maximizethe joint economicbenefits to the two countries. In terms of the present framework, this would allow us to select the weightingof the two countries' economictargets in an optimal manner. That is, through weighted voting or other means, the regional institution's objective function could be altered to the benefit of both economies. Hence we now assume that the countries could jointly commit to an institutionalsetup at the regional level, even though they cannot commit to particular policies, once the institutionsare in place. The fact that these institutionsare being set up from scratch may enable such insdtutional engineeringat the internadonal level when they are ruled out domestically.

The analysis proceeds as before, except that the game is now in three stages, with the first stage of the game consistingof choosingthe weighingscheme that determinesthe importance to be attached by the regional institutionsto each countries' bliss points. It can be shown [see de Melo,

Panagariyaand Rodrik (1992)] that the institutional-designquestion will result in putting a larger weight on the low-interventioncountry's preferences.

3.4 Political Consequences of RI

Our framework suggests that political considerationsand the constraints placed by private- sector lobbyingactivides should be taken into account in evaluating the benefits of RI. For the same reasons, political considerationsshould also weigh heavily in the determinationof when countries are likely to enter regional arrangements.

The maximizedvalues of G and G in the absence of integration are:

GN1= G =-(O 2/4)(y/(1 + y)]3, (20)

Using (14) and (16), we obtain the analogousexpression under integration: 31

2 = GI" - ([(1 + y)/4] (i - ig) + (a2/16) [y/(1 + y)]31 (21)

Considerfirst the case where the two countries are identicaland g = In this case, integration is

unambiguouslybeneficial. This derives entirely from the preference-dilutioneffect discussed above.

Integration reduces the influenceof the national lobby, and allows the governmentto get closer to

achievingits dual objectives. When g o g, on the other hand, the benefits of integration are unambiguouslyreduced for both partners. The reason is that any economic gains from preference asymmetryare offset by political losses arising from a wider gap between the national lobby's preferences (a and e) and the common level of intervention. This captures the political cost of giving up national autonomy: while economicperformance may be improved, this comes at the cost of a reduction in the government's ability to satisfy politically-importantgroups' demands. On balance, asymmetryhurts. A comparisonof (20) and (21) makes clear that regional integration would not pay for either governmentwhen g and j' are too far apart.

To summarize, regional integration entails a commitmentto abide by supra-nationalrules.

However, since these rules are endogenousand chosen by governmentesthemselves, there is no real sense in which integrationcan solve the dynamic-inconsistencyproblems that afflict policy making.

What integration can do, as this framework has highlighted,is to alter the parameters of the situationin certain ways to alleviatethe costs of dynamic-inconsistency.Hence regional integration can help governmentsachieve superior outcomes, even when integrationdoes not involve specific commitmentsto either other partners' policies or to arbitrarily selected policies. Under certain conditions,the arbitraging of institutionsacross national borders can be mutually beneficial. But regional partners have to cross a threshold of similaritybefore integration can become beneficialin this institutionalsense and, as argued by Winters in his contributionto this volume, the adoption of common policies and common rules become additionalobjectives in their own right.25 32

4. Growth Effects of RI Schemes

The institutionalconsiderations discussed above suggest that successful RI schemes benefit

from a convergencein objectives, from having few partners, and from the willingnessof countries to

surrender some national autonomyand to commit to supranationalrules. If these institutional

cknsiderationsare met, it is likely that dynamicgains, reflected in higher growth, will be reaped. We

have not referred above to the "dynamic" effects, not only because they are rarely addressed in

empirical work' but also because, like scale efficiency effects, they are not particular to RI

arrangements. However these effects, though difficult to identify, are in practice likely to be more

importantthan the static efficiency effects alluded to above. On macroeconomicgrounds "good' policy coordinationand less instabilityis likely to raise long-run growth. In past RI schemes, the benefits of coordinationthrough delegation to supra-nationalbodies was largely absent from developingcountry arrangementsbecause of a combinationof conflictingobjectives (e.g., in Africa) and weak central institutions. On the other hand, such benefits were apparent in the EC. (See the contributionsby Foroutan and Winters to this volume.)

The most often-citeddynamic effects are spillover effects and moving down learning curves.

For example, technologicaldiffusion is more likely to be rapid if increasedcompetition from trade puts pressures on domestic firms to adopt these new technologies.An enlarged market also increases the stimulus for investmentto take advantageof the enlarged market and to meet the expanding competition.Thus reduced barriers to trade would provide another strong impetus to the convergence predictionsof neoclassicalgrowth theory based on diminishingreturns. On Gershenkroniangrounds, one would argue that this catch-up effect would be stronger for developingcountries provided, of course, that few impedimentsexist on the importationof technologyfrom developed countries. Since 33 such barriers were erected rather than removed as they were part of the inward-lookingdevelopment

strategy, this dynamic effect was not exploitedto its full potential for past developing-countryRI

schemes.

Perhaps the most importantpotential dynamicbenefit for developingRI schemes comes from economic cooperationin areas where significantexternalities and public goods (education,research

and development, infrastructure,environment) exist. Of course, cooperationcan take many different forms ranging from the simple exchangeof informationthrough the provision of joint training facilities to the mutual recognitionand adoption of rules and regulations, to the implementationof joint policies and the establishmentof joint institutionswith quasi-legislativepowers. Given the general lack of institutionaldevelopment when the early RI schemes were implementedin the early sixties, it would appear that the potential for gains even from limitedregional cooperationwould be great.

Is there any evidence of positive dynamiceffects from RI? In other work [de Melo,

Montenegro, Panagariya (1992)] we look for the evidenceby fitting a simple growth equationto a cross-sectionof 101 countries stratified into a group of OECD developed(23) and developing(78) countries. We test for the eventual influenceof belonging to an RI scheme by the inclusionof a dummy variable. The model is estimated over the period 1960-85and by subperiods (1960-72 and

1973-85). Besides investment, the explanatoryvariables includeinitial per capita income and an estimate of the stock of human capital.

We tested for the effects of RI by includingdummy variables for the followingarrangements

(years of implementationin parenthesis): (i) among developed countries, EC (1960) and EFTA

(1960); (ii) among developingcountries CACM (1960), LAFrA (1960), SACU (1969) and CEAO

(1974). Among the latter group, SACU (Botswana,Lesotho, South Africa and Swaziland)is an 34 exampleof a North-SouthRI scheme and CEAO which includes former French colonies are also part of a monetaryunion since they are all members of the CFA zone. 2

Three results stand out. First, with one exception,' none of the integrationdummies was significant. Insofar as splitting the sample controls for some of the effects of omitted variables, which would capture the effects of RI, belonging to an RI had no apparent effect on long-run growth. There

is apparently no effect of membership,even for developedcountries when the sample was split into two sub-periods: 1960-72 and 1973-85. To some extent this should not come as a surprise since trade liberalizationwas being carried out multilaterallyand benefits were therefore being spread out fairly evenly. Thus there is no apparent effect of membershipin terms of higher growth even for

EC, EFTA and CACM during the 1960s when intra-regionaltrade was growing rapidly.

The second result is that splitting the sample into developedand developingcountry groups reveals an interestingdifference in the role of investmentand education in explaininggrowth.

Investmenthas the expectedpositive (and statisticallysignificant) sign for both groups and both periods, but the human capital variable, though of the right positive sign, is only significantfor the developingcountry group. This result has two interestinginterpretations. First, as emphasizedby the "new" growth literature, human capital is a contributingfactor to growth for poor countries.

Second, as emphasizedin the institutionalanalysis, and in the literature emphasiziagcooperation, there would appear to be benefits from institution-buildingand joint training. This aspect of integration, largely neglectedduring the first wave of RI arrangements, would appear to promise benefits.

Third, initial income per capita enters with a negative and statisticallysignificant sign for all regressions except for developedcountries during 1973-85. This corroborates recent findings in the growth literature [e.g. Barro (1981)]. From the point of view of RI arrangementsthis result, which 35 supports the catch-uphypothesis of neoclassicalgrowth theory, suggests that the new form of North-

South RI is promising for developingcountry members.

The first result, however, invites a note of caution in interpreting the no-effectresults on

growth. It could be argued that some of the regressors are correlated with the dummy variable

controllingfor RI membership- for example the investmentshare in GDP could be higher among RI

countriesbecause of the positive effect of membershipon the macro and institutionalenvironment. In particular, it was found that the investmentrate in the EC, and especially in EFTA, was significantly higher (about 5 percentage points) than for other developedcountries until 1973. This significantly higher investmentrate is consistent with the positive effects that one would expect from effective integrationamong institutionsand also correspondedto the period when the liberalizationpolicies were strongest and the EC's relative growth performancewas strongest. While this result is not strong evidenceof 'dynamic" effects of RI, it suggests the possibilityof an associationbetween comprehensiveRI and growth. Interestingly,the same increased investmenteffect also accompanied the announcementof NAFTA negotiationsin Mexico. [See Whalley (forthcoming)].

S. Conclusions

Where do we stand? The evidencediscussed in section 4 supports earlier findings that countries involved in RI schemes did not fare any better than comparators. But then, neither was their performance any worse. The outcome was thus quite consistent with the theoreticaldiscussion in section 2 which pointed out that conflictingfactors determine the efficiency consequencesof RI arrangements. However, history won't repeat itself as circumstancestoday are quite different from those prevailing during the first wave of RI arrangements: fewer trade restrictions, an agenda in most

RI arrangementsthat extends beyond the measures consideredunder the GATT, and a new form of 36 integrationinvolving reciprocal arrangementsbetween developedand developingpartners. On the

whole, our conclusionis that under today's circumstances,the new form of arrangementsis likely to

result in a more favorable outcomefor RI schemesprovided that real authority is delegated to well-

designedregional institutionsthat do more than respond to pressure groups, which is a big if.

What the discussionin section 3 showed, is that regional partners have to cross a threshold of

similarityin their economicobjectives before such partial surrender of national autonomy can be mutually beneficial. This has apparentlyoccurred. The political benefits of RI are maximizedby

integratingwith countries that are as similar as possible in their objectives. But from an economic perspective, as we discussed in section 3, there is an optimaldegree of dissimilarity: Mexico hopes to benefit from NAFTA preciselybecause standardsof governmentintervention and private property are much more stringent in the U.S. (and Canada) than they are in Mexico. Normally, these benefits are one-sided: continuingthe example, U.S. (and Canadian)institutions may well be contaminatedby the less advantageousMexican institutions. But as we have discussed, the economic benefits of dissimilaritycan be shared by all parties provided the common institutionsare designed in a way that weights the objective functions of each country appropriately. Hence ability to mold regional institutionsfrom scratch is an importantdeterminant of the gains from integration. However, as the experienceof the EC suggests, regional institutionsshould resist communitarisingrestrictive unilateral actions by members, and the necessarypragmatic approach to negotiatingchanges should avoid excessivecompliance with provisions catering to sectional interests.

Looking ahead, the lessons are:

(i) If the current trend of unilateralliberalization continues, integration is less likely to have

negative efficiencyeffects and can help cement hard-foughtreductions in protection. 37 (ii) By includingon their agenda areas not covered by multilateralnegotiation, such as standards

and governmentprocurement, RI arrangementscan serve as a catalyst in the pursuit of

removing market fragmentations.

(iii) Integrationcan confer substantialinstitutional benefits, but only if real authority is delegated

to central institutions. As long as multilateralinstitutions are weak, and regional

arrangementsallow for a greater surrender of national autonomy, regionalismmay

paradoxicallyremain attractiveto reforming governments. However, strengthened

multilateralinstitutions would provide greater benefits than regional ones as they will present

a larger political communityand a greater scope for the preference-dilutioneffect.

(iv) To benefit from integration, some convergencein national economicobjectives is necessary.

This explains why the prospect of integrationwith rich neighbors was so unattractiveto

developingcountries when they still believed in import substitution, and whiyit has become

more desirableto do so as they have opened up their economies. It also explains why

multilateralinstitutions may be inherentlymore difficult to sustain than regional institutions

in light of the underlying diversity of national interests.

(v) However, too much convergencemay undercut the economicgains from integration. This

explains why developing countries currently stand to benefit least (in economicterms, at

least) by integratingwith other developingcountries with similarly unsettled domestic

institutions.

(vi) The benefits of integrationare greater the higher the possibilityof shaping the institutions

that go along with it in an economicallydesirable way. Recent regional attempts perhaps

offer an additional degree of freedom in this regard comparedto existing multilateral

institutions. 38 Endnotes

1. This means that transshipmentrules are necessary to prevent imports of high-tariffcountries from being channeled via low-tariff countries in the area.

2. In addition to the literature cited later, importantcontributions to the literature includeCorden (1968, 1970, 1972, 1976), Kemp (1969), Lipsey (1957, 1960, 1970), Riezman (1979) and Vanek (1964). Two excellent surveys of the literature are Corden (1984) and Lloyd (1979).

3. Accordingto Viner, trade creation occurs when the formationof an FTA or CU leads to a switching of imports from a high-costsource to a low-cost source. Analogously,trade diversion occurs when imports switch from a low-costsource to a high-cost source. Subsequentauthors noted crucial limitationsof his conclusionand demonstratedthat in general trade diversion need not be welfare worsening and trade creation need not be welfare improving. The concepts have remained highly influentialin policy debates, however, and authors have continuedto equate trade creation with welfare improvementand trade diversion with welfare deterioration.

4. In terms of Viner's terminology,the FTA just described is trade diverting. Yet the net effect of it on welfare is ambiguous. This is the point made in the post-Viner literature alluded to in footnote 2.

5. As we will argue later, this is easier said than done. In practice, compensationschemes can turn even a welfare enhancingFTA into a disaster.

6. Shortly we will see that once we recognizethat the imports from B and C are impefect substitutes, the positive and negative effects emerge without eliminationof trade with C. Figure 1 also provides a less extreme representationof the standard Vinerian analysis in which the supply curves of both B and C are horizontal.

7. In the case described in the previous paragraph, the FTA is unambiguouslywelfare worsening. Therefore, it is not of much interest from a policy perspective.

8. This was the issue raised Cooper and Massell (1965a) and Johnson (1965) and also discussed by Berglas (1979) and Robson (1980).

9. The following discussiondraws heavilyupon Wonnacottand Wonnacott (1981). Also see Berglas (1983) and Wonnacott and Wonnacott(1984) in this context.

10. By the Lerner SymmetryTheorem, the tariff on good 2 by B is equivalentto an export tax on good 1 at the same rate. Therefore, the tariff in B impliesan upward shift in its export supply curve.

11. As Wonnacott and Wonnacott(1981) show, eliminationof trade with C is not necessaryto generate this result. What is critical is that the good imported by C before the FTA be different than that after it.

12. Although we did not consider formallythe option of accessationto a bloc but it is clear from Figure 2 that if the internal price of C is above point G, B will be better off acceding to C than forming an FTA with A. 39 13. This was the main thrust of the analyses in Cooper and Massell (1965b) and Johnson (1965).

14. See Bhagwati (1968, 1990) in this context.

15. Strictly speaking, the externalityimplies that if the equilibriumis incompletelyspecialized, some interventionwill be desirable and complete free trade will not be locally optimal. In the discussionjust presented, we have ignored this complication.

16. The literature on imperfect competitionand FTAs is almost nonexistent. Therefore, the best we can do is infer such implicationsfrom the models addressingnational trade policies in the presence of imperfect competition. Unfortunately,even here the results depend crucially on the specificstructure of the model.

17. For examplesof monopolisticcompetition, see Helpman (1981) and Krugman (1980).

18. This model is due to Meade (1955).

19. In models with three or more goods, the terms of trade effects are more complicatedand less clear cut, although the presumption is that the union's terms of trade will improve upon the formation of an FTA. Thus, in Meade's model, as each partner country reduces its tariff on imports from the other, givea substitutability,the demand for imports from the outside world declines. The union's terms of trade with respect to the outside world improve.

20. This is, of course, a scenario in which FTAs serve as stepping stones to multilateralfree trade.

21. The importanceof gaining market access is also stressed by Winters in his discussionof successiveenlargements of the EC in his contributionto this volume.

22. "NAFTA is importantto Mexico in itself.... But it is still more important as a symbolic guaranteeof the nature of the changes in the Mexican economy. Since the mid-1980s, when President Miguel de la Madrid took Mexico into GATT and slashed tariffs, its technocratshave used one word to describe the changes: 'irreversible'. A freetrade agreement would, they hoped, prove the point. "The hard decisions ... were taken before Mr. Salinas was publicly convincedof the case for free trade with Canada and the United States. But by signallingthat within ten years he wants borderless trade within North America, he is ellingboth Mexico's once inefficientindustries and the outside world that there really is no going back." (Tle Economis, December 14, 1991, p. 20.)

23. That is because the simplesum of the objective functionsdiffers from expression (9) only by terms that are constant from the perspectiveof the regional institution. Consequently,the first-order conditionsare identical in the two cases.

24. A couple of things could work against this in practice. First, political groups may form transnational alliances. But even if they do, larger groups will likely suffer more from free-rider problems and from diversity of interests. Second, regional institutionsmay somehowturn out to be more receptive to pressure than the national institutions. This problem can be alleviatedby careful institutionaldesign. See below. 40 25. The advantageof a relatively small number of participants in a world of trading blocs Is also made by Mancur Olson in his commentson Douglas Irwin's paper in this volume.

26. On the difficult role and mixed outcome of supra-nationalinstitutions to help establisha commonmarket in the EC, see section 4 of Winters' contributionto this volume.

27. Often-citedstudies of the dynamiceffects are Cline (1978) for the CACM and Baldwin (1991) for the effects of 1992on EC growth.

28. Other RI schemes that were not includedeither did not have a long enough existence (e. g. SADC, Canada-U.S. FTA, etc.) or did not achieve much by way of integration.

29. The only exception is the significanteffect for the dummy variable for SACU (which is entered for Botswana,Lesotho and Swaziland)for the period 1960-72. SACU is a highly integrated RI in existence since 1910 and the members' relation to South Africa is much like that of Liechtensteinto Austria or Monacoto France. 41 References

Barro, R. (1991), "EconomicGrowth in a Cross-Sectionof Countries", Quarterly Journal ot Economics.

Berglas, E. (1983), "The Case for Unilateral Tariff Reductions: Foreign Tariffs Rediscovered", American EconomicReview, vol. 73, pp. 1141-112.

Berglas, E. (1979), "Preferentialtrading theory: The n commoditycase", Journal of Political Economv, 87:315-331.

Bhagwati, J. (1990), "Departuresfrom Multilateralism: Regionalismand Aggressive Unilateralism", The EconomicJournal, 100: 1304-1317.

Bhagwati, J. (1968), "Trade LiberalizationAmong LDCs, Trade Theory and GATT Rules". In Yalue. Capital and Growth: Papers in Honor of Sir John Hicks. Oxford University Press.

Chenery, H. and M. Syrquin (1975), Patterns of Development:1950-1970 Oxford: Oxford University Press.

Cline, W.R. (1978), "Benefitsand Costs of Economic Integrationin Central America", in W.R. Cline and C. Delgado eds. EconomicIntegration in Central America, Washington, D.C. Brookings Institution.

Cooper, C. A. and B. F. Massell (1965a), "A new look at customsunion theory", Economic Journal, 75:742-747.

Cooper, C. A. and B. F. Massell (1965b), "Toward a general theory of customs unions for developing countries", Journal of Political Economy, 73:461 476.

Corden, W. M. (1984), "NormativeTheory of InternationalTrade", ch. 2 of R. W. Jones and P. B. Kenen (eds.), Handbook of InternationalEconomics, North-Holland,Amsterdam.

Corden, W. M. (1976), 'Customs union theory and nonuniformityof tariffs", Journal of International Economics,6:99-106.

Corden, W. M. (1972), "Economiesof scale and customsunion theory", Journal of Political Econom, 80:465475.

Corden, W. M. (1970), "The efficiency effects of trade and protection", in I. A. McDougall and R. H. Snape, eds., Studies in InternationalEconomics: Monash Conferencepapers (North- Holland, Amsterdam).

Corden, W. M. (1968), CustomsUnion Theory, unpublishedmanuscript. 42 Eaton, J. and G. M. Grossman (1986), "OptimalTrade and Industrial Policy Under Oligopoly", Ouarterly Journal of Economics,pp. 383-406.

El-Agraa, A.M. ed. (1988), "The European Community"in A.M. El-Agraa ed. International Economic Integration, London, MacMillan.

Erzan, R., and A. Yeats (1992), "Free Trade Agreements and the United States: What's in it for Latin America?" PRE working paper no 827, Washington, D.C.,World Bank.

Fischer, S. (1990), "Rules versus Discretion in Monetary Policy", in B. M. Friedman and F. H. Hahn (eds.), Handbook of MonetaryEconomics, vol. II, Amsterdam, North-Holland.

Foroutan, F. (forthcoming),"Regional Integration in Sub-SaharanAfrica: Past Experience and Future Prospects", in J. de Melo and A. Panagariyaeds., New Dimensions in Regional Integration (this volume).

Greenaway,D. (1989), "RegionalTrading Arrangementsand Intra-Industry Schott ed. Trade: Evidence and Policy Issues" in D. Greenawayet al. eds. Economic Aspects of Regional Trading Arrangements,New York: New York UniversityPress.

Hamilton, A., J. Madison, and J. Jay, 1981[1787], The Federalist Pagers, ed. Roy P. Fairfield, The Johns Hopkins University Press, 2nd ed., Baltimoreand London.

Helpman, E. (1981), "InternationalTrade in the Presence of Product Differentiation,Economies of Scale and MonopolisticCompetition: A Chamberlin-Heckscher-OhlinApproach', Journal of InternationalEconomics, 11(3): 305-40.

Hirschman, A. (1971), A Bias for Hope. New Haven and London, Yale University Press.

Irwin, D. (forthcoming), "Multilateraland BilateralTrade Policies in the World Trading System: An Historical Perspective",in J. de Melo and A. Panagariyaeds., New Dimensionsin RegionalInteeration.

Johnson, H. G. (1965), "An economictheory of protectionism,tariff bargaining, and the formation of customs unions", Journal of PoliticalEconomy, 73:256-283.

Kemp, M. C. (1969), "A contributionto the general equilibriumtheory of preferential trading", (North-Holland,Amsterdam).

Kowalczyk, C. (1990), "Welfare and Customs Unions", N.B.E.R. Working Paper No. 3476.

Krugman, P. (1984), "U.S. Responseto Foreign Industrial Targeting", BrookingsPapers on Economic Activity, no. 1:77-131.

Krugman, P. (1980), "Scale Economies,Product Differentiation,and the Pattern of Trade", Americn EconomicReview, vol. 70, pp. 950-959. 43 Lachler, U. (1989), 'Regional Integrationand EconomicDevelopment", Industry and r..ergy Department Working Paper no. 14.

Langhammer,R. and U. Hiemenz (1992), "RegionalIntegration among DevelopingCountries: Survey of Past Performanceand Agenda for Future Policy Action", UNDP-WorldBank Trade ExpansionProgram OccasionalPaper 7, Trade Policy Division, The World Bank, Washington, D.C.

Langhammer,R. and U. Hiemenz (1990), "RegionalIntegration among DevelopingCountries- Opportunities, Obstacles and Options", Kieler Studien 232, Tubingen: J.C.B. Mohr.

Lipsey, R. G. (1970), "The theory of customsunions: A general equilibriumanalysis", (Weidenfeld and Nicolson, London).

Lipsey, R. G. (1957), "The theory of customsunions: Trade diversion and welfare", Economica, 24:40-46.

Lipsey, R. G. (1960), "The theory of customsunions: A general survey", Economic Journal, 70:498-513.

Lloyd, P. J. (1982), "3 x 3 theory of customs unions", Journal of InternationalEconomics, 12:41-63.

Lopez, R. and A. Panagariya (1992), "On the Theory of PiecemealTariff Reform: The Case of Pure Imported IntermediateInputs", AmericanEconomic Review, June, forthcoming.

Meade, J. E. (1955), The theory of customsunions (North-Holland,Amsterdam).

Melo, J. de, C. Montenegro and A. Panagariya(1992), "RegionalIntegration Old and New: Issues and Evidence" (mimneo),World Bank.

Melo, J. de, A. Panagariya and D. Rodrik (1992), "RegionalIntegration: An Analytical and Empirical Overview" (mimeo), World Bank.

Nogues, J. and R. Quintanilla (forthcoming),"Latin America's Integration and the Multilateral Trading System" in J. de Melo and A. Panagariyaeds., New Dimensionsin Regional Integration.

Olson, Mancur (1992) "Multilateraland BilateralTrade Policies in the World Trading System" (mimeo), World Bank.

Riezman, R. (1979), "A 3 x 3 model of customs unions", Journal of InternationalEconomics, 72:820-829.

Robson, P. (1980), The economicsof internationalintegration, G. Allen and Irwin, London, Boston.

Schott, J. (1988), ed. Free Trade Areas and U.S.Trade Policy Washington,D.C., International EconomicInstitute. 44 Summers,R. and A. Heston, (1988), "A New Set of InternationalComparisons of Real Product Prices and Price Levels: Estimates for 130 Countries", Review of Income and Wealth.

Vanek, J. (1964), "Unilateral trade liberalizationand global world income", Quarterly Journal of Economics, 78:139-47.

Viner, J. (1950), The customs union issue (CarnegieEndowment for InternationalPeace, New York).

Whalley, J. (forthcoming),"Regional Trade Arrangementsin North America: CUSTA and NAFTA" in J. de Melo and A. Panagariyaeds., New Dimensionsin Regional Integration.

Winters, A. (1993), "The European Community: A Case of SuccessfulIntegration? New Dimensionsin Regional Integration(this volume).

Wonnacott,P. and M. Lutz (1988) "Is There a Case for Free Trade Areas", In J. Schott ed.

Wonnacott,P. and R. Wonnacott(1984), "How General is the Case for Unilateral Tariff Reduction?" AmericanEconomic Review, 74:491.

Wonnacott, P. and R. Wonnacott(1981), "Is unilateraltariff reduction preferable to a customsunion? The curious case of the missing foreign tariffs", AmericanEconomic Review, 71:704-714.

Wooton, I (1988), "Towards a Common Market: Factor Mobilityin a Customs Union", Canadian Journal of Economics, pp. 525-538. Figure 1: Basic Analysis of an FTA. Relative Price

M1A A's gain: Horizontally shaded triangle less vertically shaded triangle.

MlA

E1 B

(1 t)pc =PA D ___

pc R U|p

M1A

- I mlA'

Quantity imported Figure 2: FTAunder a QR Relative Price

B's gain: SFRPC A's gain: Horizontally shaded E1B triangle less vertically shaded rectangle.

S%

GvL H G 4%~~~~~4 N5l71iilTJ>s

ml A

ImlrA'

Quantity imported Price Figure 3: FTA vs. UTR

M A El18

M1 AQa pcL MN R

A' 1A

ElB mlA'

Quantity Pollcy Research Working Paper Series

Contact Title Author Date for paper

WPS1072Costs of AlternativeTreatments for BrookeR. Johnson January1993 0. Nadora IncompleteAbortion JanieBenson 31091 Janet Bradley AuroraRabago Ordoflez CatiaZambrano LeonardOkoko LeticiaVazquez Chavez PaulinaQuiroz KhamaRogo

WPS1073Fiscal Aspects of Developing PeterJ. Montiel January1993 R. Vo CountryDebt Problems and Debt-Service 31047 ReductionOperations: A Conceptual Framework

WPS1074 HowMoving to WorldPrices Affects DavidG. Tarr January1993 N. Artis theTerms of Tradein 15Countries 38010 of theFormer Soviet Union

WPS1075Many Paths to SkilledEmployment: AdrianZiderman January1993 PHREE A ReverseTracer Study of Eight RobinHorn 33680 Occupationsin Colombia

WPS1076A PresumptivePigouvian Tax on GunnarS. Eskeland January1993 C. Jones Gasoline:Analysis of An AirPollution 37699 ControlProgram for MexicoCity

WPS1077Risk Management Prospects for PanosVarangis January1993 D. Gustafson EgyptianCotton EltonThigpen 33714 TakamasaAkiyama

WPS1078The Evolutionof Welfareand Poverty Christiaan Grootaert January1993 E. Vitanov UnderStructural Change and Economic 38400 Recessionin C6ted'lvoire, 1985-88

WPS1079How Useful Are Integrated Household Christiaan Grootaert January1993 E. Vitanov SurveyData for Policy-Oriented 38400 Analysesof Poverty?Lessons from the C6ted'lvoire Living Standards Survey

WPS1080A NewRegional Price Index for C6te ChristiaanGrootaert January1993 E. Vitanov d'lvoireUsing Data from the RaviKanbur 38400 InternationalComparisons Project

WPS1081Correcting for SamplingBias in the LionelDemery January1993 E. Vitanov Measurementof WeHfareand ChristiaanGrootaert 38400 Poverty:The Case of theC6te d'lvoire UvingStandards Survey Pollcy Research Working Paper Series

Contact Title Author Date for paper

WPS1082What Do GoveMments Buy? The ShantayananDevarajan February1993 C. Jones Compositionof PublicSpending and VinayaSwaroop 37699 EconomicPerformance Heng-fuZou

WPS1083Finance and Growth: Schumpeter RobertG. King February1993 D. Evans MightBe Right RossLevine 38526

WPS1084Stock Market Development and DongHe February1993 P. Infante FinancialDeepening in Developing RobertPardy 37665 Countries:Some Correlation Patterns

WPS1085 EconomicApproaches to Modeling CristinoR. ArroyoliI February1993 0. Nadora FertilityDeterminants: A Selective Review 31091

WPS1086Teachers' Salaries in LatinAmerica: GeorgePsacharopoulos February1993 G. Psacharopoulos A ComparativeAnalysis JorgeValenzuela 39243 MaryArends

WPS1087Exchange-Rate-Based Stabilization: Alberto F. Ades February1993 R. Luz Talesfrom Europe and Latin America MiguelA. Kiguel 34303 NissanLiviatan

WPS1088A Primeron the MFA Maze RiccardoFaini February1993 D. Ballantyne Jaimede Melo 37947 WendyTakacs

WPS1089Equity Portfolio Investment in StijnClaessens February1993 R. Vo DevelopingCountries: A LiteratureSurvey 31047

WPS1090 GovernmentExpenditures as a ThanosCatsambas February1993 A. Correa Citizen'sEvaluation of PublicOutput: 38549 PublicChoice and the BenefitPrinciple of Taxation

WPS1091 CapitalMarket Imperfections Before FidelJaramillo February1993 W. Pitayatonakarn andAfter Financial Liberalization: FabioSchiantarelli 37658 A Euler-EquationApproach to Panel AndrewWeiss Datafor EcuadorianFirms

WPS1092The Effect of FinancialLiberalization Fidel Jaramillo February1993 W. Pitayatonakarn on theAllocation of Credit:Panel FabioSchiantarelli 37658 DataEvidence for Ecuador AndrewWeiss

WPS1093Swiss Chilanpore: The Way Forward DimitriVittas February1993 W. Pitayatonakarn for PensionReform? 37658

WPS1094The New Regionalism: A Country Jaimede Melo February1993 D. Ballantyne Perspective ArvindPanagariya 37947 DaniRodrik